I have analyzed the NHC official forecast error database. However, there is a significant inconsistency among the dataset, Official error trends, and annual report.
Surprisingly, there are the same record, 11518 rows including header in Official NHC Atlantic Track and Intensity Forecast Errors (1970-2019, excluding depressions) and Official NHC Atlantic Track and Intensity Forecast Errors (1989-2019, including depressions). I cannot believe that the data with different temporal range with different criterion have the same records!
Take 2019 official track forecast errors of 48h as example. The annual report says there are 187 records. The official error trends table says there are 213 records and the average is 74.7 n mi. However, the data says there are only 166 records and the average is 76.11 n mi. Including TDs or not has no effect on the results.
The inconsistence exits in the results of other years.